So if I understand correctly, it is only that specific VM that is running on one Hyper-V host. All the other VMs on the two other hosts are still getting the normal speed. Correct?
Since VBR states that the bottleneck is the network, I would believe something has changed over there. So can you look at the following:
* Any windows installed (exchange server/ hyper-v host) that are not installed on the other servers (or vice versa)
* Difference in firmware for the NIC on the physical box?
* Can you use IPERF or Wireshark to test the speed between the VBR box and the VM?
* Nothing else that has changed just before you started seeing the issue?
If the output of IPERF or Wireshark is normal, then please log a support call. Maybe our engineers can figure out what is exactly wrong. We had many cases where an update, wrong NIC driver/ firmware or similar was causing the issue